[gradsusr] help with the ensemble dimension in control file

[Jeff Duda]

Hello,
I have two WRF ensembles (both 7 members).  In one ensemble, only the value
of a constant in the dynamics section varies among the members (ensemble
1).  In the other ensemble, the choice of microphysics, PBL, and land
surface models vary (ensemble 2).  All simulations are on the exact same
grid with the same history dump interval.  The control files I'm using to
view the ensembles are shown below.

FOR ENSEMBLE 1
---------------------------------
dset /scratch/jdduda/ensemble_1/%e/wrfout_d01_%y4-%m2-%d2_%h2:00:00
dtype netcdf
undef -9999.0
options template
PDEF 599 549 lcc 25.7907 -108.7018 1 1 38.5 38.5 -97 4000 4000
XDEF 599 linear -112.2245 0.050833
YDEF 549 linear 25.79018 0.036935
ZDEF 39 linear 1 1
TDEF 25 linear 12Z13APR2012 1hr
EDEF 7 names m1 m2 m3 m4 m5 m6 m7
VARS 5
P=>p 39 t,z,y,x perturbation pressure (Pa)
PB=>pb 39 t,z,y,x base state pressure (Pa)
T=>temp 39 t,z,y,x theta-290 K (K)
RT_TENDF_STOCH=>utend 39 t,z,y,x temperature perturbation from SKEB (K)
RAINNC=>rainnc 1 t,y,x precipitation (mm)
ENDVARS

FOR ENSEMBLE 2
--------------------------------
dset /scratch/jdduda/ensemble_2/%e/wrfout_d01_%y4-%m2-%d2_%h2:00:00
dtype netcdf
undef -9999.0
options template
PDEF 599 549 lcc 25.7907 -108.7018 1 1 38.5 38.5 -97 4000 4000
XDEF 599 linear -112.2245 0.050833
YDEF 549 linear 25.79018 0.036935
ZDEF 39 linear 1 1
TDEF 25 linear 12Z13APR2012 1hr
EDEF 7 names m1 m2 m3 m4 m5 m6 m7
VARS 5
P=>p 39 t,z,y,x perturbation pressure (Pa)
PB=>pb 39 t,z,y,x base state pressure (Pa)
T=>temp 39 t,z,y,x theta-290 K (K)
RT_TENDF_STOCH=>utend 39 t,z,y,x temperature perturbation from SKEB (K)
RAINNC=>rainnc 1 t,y,x precipitation (mm)
ENDVARS

The difference between them is essentially one character in the dset entry
("1" vs. "2")

When I attempt to run a script to plot a time series of ensemble mean and
spread of the variable RAINNC, it works fine for ensemble 1.  However, for
ensemble 2, I get this error:

NetCDF Error (gancrow, nc_get_vara_double): NetCDF: Index exceeds dimension
bound

Looking further, I see that the history files indeed differ between the
ensemble members of ensemble 2.  They differ in the number and type of
arrays contained.  However, each member has each of the 5 variables
indicated in the control file.  The trouble seems to especially come from
member m3.  Despite the fact that member m3 differs from the other members
in a similar way as the other members differ from each other, I cannot see
why Grads would be giving me this problem.  I can use ncview to view the
RAINNC variable for m3 as well as for other members alone.  Why am I having
this problem when using the ensemble dimension to template in a control
file?

Jeff Duda

-- 
Jeff Duda
Graduate research assistant
University of Oklahoma School of Meteorology
Center for Analysis and Prediction of Storms
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gradsusr.org/pipermail/gradsusr/attachments/20130422/1406c451/attachment-0003.html